Code Excited Linear Prediction Speech Decoder and Method Thereof

1. A code excited linear prediction speech decoder, comprising: an adaptive codebook configured to generate an adaptive code vector; a random codebook configured to generate a random code vector, the random codebook comprising: an input vector providing section configured to provide an input vector comprising at least one pulse, each pulse having a predetermined position and a respective polarity, a fixed waveform storage section configured to store at least one fixed waveform, a determination section configured to determine whether input speech is at least one of voiced and unvoiced, and to select the at least one fixed waveform when the input speech is determined to be unvoiced, while producing the input vector as the random code vector when the input speech is determined to be voiced, and a convolution section configured to convolute the at least one fixed waveform, which is selected when the input speech is determined to be unvoiced, with the input vector to thereby produce the random code vector; and a synthesis filter that receives a signal based on the adaptive code vector and the random code vector, and is configured to perfonn linear prediction coefficient synthesis on the signal to output synthesized speech.

2. The code excited linear prediction speech decoder of claim 1 , wherein the fixed waveform storage section includes at least two fixed waveforms, the determination section selects one of the at least two fixed waveforms when the input speech is determined to be unvoiced, and the convolution section convolutes one of the at least two fixed waveforms, which is selected when the input speech is determined to be unvoiced, with the input vector to thereby produce the random code vector.

3. The code excited linear prediction speech decoder of claim 1 , wherein the determination section determines whether input speech is at least one of voiced and unvoiced based on a value of a quantized pitch gain.

4. The code excited linear prediction speech decoder of claim 1 , wherein the determination section determines whether input speech is at least one of voiced and unvoiced based on an adaptive codebook gain.

5. A method of producing synthesized speech, comprising: storing at least one fixed waveform; providing an adaptive code vector; determining whether input speech is at least one of voiced and unvoiced; selecting the at least one fixed waveform when the input speech is determined to be unvoiced; providing an input vector comprising at least one pulse, each pulse having a predetermined position and a respective polarity; producing the input vector as a random code vector when the input speech is determined to be voiced; convoluting the at least one fixed waveform, which is selected when the input speech is determined to be unvoiced, with the input vector to produce a random code vector; and performing linear prediction coefficient synthesis on a signal based on the adaptive code vector and the random code vector to output synthesized speech.

6. The method of claim 5 , wherein the storing at least one fixed waveform stores at least two fixed waveforms, the selecting the at least one fixed waveform selects one of the at least two fixed waveforms when the input speech is determined to be unvoiced, and the convoluting the at least one fixed waveform convolutes one of the at least two fixed waveforms, which is selected when the input speech is determined to be unvoiced, with the input vector to thereby produce the random code vector.

7. The method of claim 5 , wherein the determining whether input speech is at least one of voiced and unvoiced is performed based on a value of a quantized pitch gain.

8. The method of claim 5 , wherein the determining whether input speech is at least one of voiced and unvoiced is performed based on an adaptive codebook gain.

9. A code excited linear prediction speech decoder, comprising: an adaptive codebook configured to generate an adaptive code vector; a random codebook configured to generate a random code vector, the random codebook comprising: an input vector providing section configured to provide an input vector comprising at least one pulse, each pulse having a predetermined position and a respective polarity, a fixed waveform storage section configured to store a first fixed waveform and a second fixed waveform, a determination section configured to categorize input speech with reference to a value of a gain, and to select the first fixed waveform or the second fixed waveform based on a result of the categorization, and a convolution section configured to convolute the selected first or second fixed waveform with the input vector in accordance with the position and the polarity of each pulse; and a synthesis filter that receives a signal based on the adaptive code vector and the random code vector, and is configured to perform linear prediction coefficient synthesis on the signal to output synthesized speech.

10. The code excited linear prediction speech decoder of claim 9 , wherein the determination section categorizes input speech based on a value of a quantized pitch gain.

11. The code excited linear prediction speech decoder of claim 9 , wherein the determination section categorizes input speech based on an adaptive codebook gain.

12. The code excited linear prediction speech decoder of claim 9 , wherein the determination section categorizes input speech into at least one of voiced and unvoiced.

13. A code excited linear prediction speech decoder, comprising: an adaptive codebook configured to generate an adaptive code vector; a random codebook configured to generate a random code vector, the random codebook comprising: an input vector providing section configured to provide an input vector comprising at least one pulse, each pulse having a predetermined position and a respective polarity, a fixed waveform storage section configured to store a first fixed waveform and a second fixed waveform, a determination section configured to categorize input speech with reference to a value of a gain, and to select the first fixed waveform or the second fixed waveform based on a result of the categorization, and a fixed waveform arranging section configured to arrange the selected first or second fixed waveform in accordance with the position and the polarity of each pulse of the input vector; and a synthesis filter that receives a signal based on the adaptive code vector and the random code vector, and is configured to perform linear prediction coefficient synthesis on the signal to output synthesized speech.

14. The code excited linear prediction speech decoder of claim 13 , wherein the determination section categorizes input speech based on a value of a quantized pitch gain.

15. The code excited linear prediction speech decoder of claim 13 , wherein the determination section categorizes input speech based on an adaptive codebook gain.

16. The code excited linear prediction speech decoder of claim 13 , wherein determination section categorizes input speech into at least one of voiced and unvoiced.

17. A method of producing synthesized speech, comprising: storing a first fixed waveform and a second fixed waveform; providing an adaptive code vector; categorizing input speech with reference to a value of a gain; selecting the first fixed waveform or the second fixed waveform based on a result of the categorization; providing an input vector comprising at least one pulse, each pulse having a predetermined position and a respective polarity; convoluting the selected first or second fixed waveform with the input vector to produce a random code vector; and performing linear prediction coefficient synthesis on a signal based on the adaptive code vector and the random code vector to output synthesized speech.

18. The method of claim 17 , wherein the categorizing input speech is performed with reference to a value of a quantized pitch gain.

19. The method of claim 17 , wherein the categorizing input speech is performed with reference to an adaptive codebook gain.

20. The method of claim 17 , wherein the categorizing input speech categorizes input speech into at least one of voiced and unvoiced.